-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
How much distortion can you hear?
- Thread starter VintageFlanker
- Start date
The "music" has fallen apart by 10%
The tone, by 2%.
Can't on/off the distortion, so those are gross (where it's become obvious) values.
The tone, by 2%.
Can't on/off the distortion, so those are gross (where it's become obvious) values.
See this thread for a more detailed listening test by Klippel.
- Joined
- Feb 23, 2016
- Messages
- 20,747
- Likes
- 37,562
Or use Pkane's Distort software for your own customized settings. Also you get to use your own choice of music.
https://distortaudio.org/
https://distortaudio.org/
Last edited:
weasels
Senior Member
fun for a quick test.
My results were similar to Ray - around 8% for the music, around 1.5% for the tone. By 10% and 2% the music has fallen apart and the tone is in harmony with itself.
My results were similar to Ray - around 8% for the music, around 1.5% for the tone. By 10% and 2% the music has fallen apart and the tone is in harmony with itself.
weasels
Senior Member
I think the electronic music makes it easier to hear the distortion. The first time I ran through in the living room with my wife watching TV next to me while I listened with open back headphones.
I repeated it in my office and had slightly different but similar results - about 7% and 1.5%. I did the Klippel test and my results were a bit better than average on the full T. Chapman track, so no super ears here. I'm also an untrained listener and in my late 40s.
I repeated it in my office and had slightly different but similar results - about 7% and 1.5%. I did the Klippel test and my results were a bit better than average on the full T. Chapman track, so no super ears here. I'm also an untrained listener and in my late 40s.
- Joined
- Oct 11, 2018
- Messages
- 3,740
- Likes
- 6,453
Most electrical gear is transparent these days. That doesn't mean one shouldn't buy good engineering. Loudspeakers have a lot of distortion still. Someone needs to work on that. Any analog source (records or tape) is a crap shoot, depending on the time of day and your luck.
I recall in the '70s when Bob Carver conducted a test similar to this (I think he was using crossover distortion) with some Stereo Review editors. Their findings were about at this level. Next issue of Stereophile (could have been a year later given his publishing schedule) Gordon Holt went off on the Carver tests for the usual idiotic reasons--didn't use 'revealing' gear, didn't use good recordings, didn't use the proper music et al. Of course back then we were all idiots when it came to audio. Some of us grew out of our idiocy, others never did, nor ever will.
Either you can hear it or you can't. If you can you can demonstrate it reliably under certain conditions. It's that simple.
Any results from that test are null and void due to a 'tell' at both the beginning and end of the test track. It's also, like the one in this thread, an online test run in a browser, so Windows shared audio mode is likely being used by most users with possible sample rate conversion. YouTube also lossily compresses audio to a maximum of 192kbps. Basically these online tests aren't really a reliable indicator of distortion audibility. You need to do a proper ABX test reaching at least 13/16 trials passed in bit-perfect WASAPI exclusive audio mode using e.g. Foobar2000's ABX comparator component to get reliable results.
Good to know.
- Joined
- Oct 11, 2018
- Messages
- 3,740
- Likes
- 6,453
Reliability means consistency. There is no reason these sorts of Internet tests cannot provide consistency. As long as one doesn't obtain wildly variant results each time one takes the test then it can be used as a valuable self screening tool. Obviously it's not going to satisfy the validity level a more rigorous control is going to offer.
To say any results obtained are 'null and void' is to overstate the case quite a bit, given what these 'tests' offer to do.
There are two definitions of reliability - the technical scientific definition, and the commonly used definition that I was using (from the Oxford English dictionary):
That's the beauty (and frustration) of the English language for you - many words have several different meanings, some even contradictory to each other. The bottom line is these tests are poor ways of determining your distortion audibility threshold, that can lead to mistruths being proliferated all over the internet. If you're going to do something, do it properly.
Last edited:
I read recently that acoustic, classical, etc music is actually a terrible way to evaluate equipment, as it is the easiest to reproduce. In contrast, pop, electronic, and female vocals reveal equipment deficiencies much more readily due to their wider freq spectrum.
- Joined
- Oct 11, 2018
- Messages
- 3,740
- Likes
- 6,453
My score on the Klippel test was average (mixed test tones). They key for me was to try and focus on something familiar in the tones, and then attempt to compare. Once you reach your threshold the guessing begins. With practice I suppose one could improve their score--at least to a specified limit when true inaudibility would manifest. I think that with complex music you'd just get lost easily, unless the distortion levels were really high. I suspect that what most people find out from taking the test is that their hearing is not as fine as they would like to think it is.
weasels
Senior Member
I can understand the sentiment, but in my opinion it's a mistake to ignore something of utility because it could be done better. To me the question isn't "is this test as accurate as possible", the question is "is this test useful?" and within it's constraints, yes it is useful.
No data is better than incorrect data that leads to false conclusions. There are just too many unquantifiable nuisance variables in these tests for them to have any trustworthy utility. It's really not complicated to just do a proper, statistically valid ABX test with Foobar anyway.
Last edited:
after covid and working for companies which specialize in data acquisition, I begin to question if there is such a thing as good data!
Trouble Maker
Addicted to Fun and Learning
This is just on a basic laptop (no external amp or DAC) with Bose QC25 noise canceling off.
Music is hard to tell exactly where it happens, it goes through the area where I feel I can start to hear it a little quickly, but maybe starting aorund 5-6% and by 7-8% it's bad.
The 500hz I could hear it around 0.008%. Around 0.4% my ears were on fire (I took my headphones off at one point). But then it seemed to subside as the distortion became more broad band... pseudo pink noise?
The 100hz I could hear around 0.02%, but it never became as annoying at any point as the 500hz.
With the tones it's not a distortion you will hear first but the warbling of sound.
I did it multiple times with my eyes closed and hit the space bar to pause when I could hear it to check with similar results.
Now I really want to do the other one when I have time.
Music is hard to tell exactly where it happens, it goes through the area where I feel I can start to hear it a little quickly, but maybe starting aorund 5-6% and by 7-8% it's bad.
The 500hz I could hear it around 0.008%. Around 0.4% my ears were on fire (I took my headphones off at one point). But then it seemed to subside as the distortion became more broad band... pseudo pink noise?
The 100hz I could hear around 0.02%, but it never became as annoying at any point as the 500hz.
With the tones it's not a distortion you will hear first but the warbling of sound.
I did it multiple times with my eyes closed and hit the space bar to pause when I could hear it to check with similar results.
Now I really want to do the other one when I have time.
Similar threads
